Incremental Inference of Black-Box Components to Support Integration Testing

Model based testing relies on the availability of formal models that are indispensable in analyzing the complete system's behavior and testing of the key functionalities. On the other hand, the system designers of the industry are mostly relying on the integration of readymade software components (COTS) to build complex applications e.g. telecom services. Unfortunately, they are not provided with formal models or with reasonable documentation. The spirit of the thesis is to devise techniques to build formal models of black-box components, and to adopt test-generation strategies based upon the learned models to support integration testing. As a first step, we are modifying previous learning method, i.e., Angluin's algorithm to work with (extended) FSM models, which incorporates input/output parameters and predicates as well. Our framework from telecom industry focuses on methods that will be applied to large-scale components. That's why we are gradually moving towards richer models such as EFSMs